Causal relationship between inflammatory cytokines and autoimmune thyroid disease: a bidirectional two-sample Mendelian randomization analysis

Background Autoimmune thyroid disease (AITD) ranks among the most prevalent thyroid diseases, with inflammatory cytokines playing a decisive role in its pathophysiological process. However, the causal relationship between the inflammatory cytokines and AITD remains elusive. Methods A two-sample Mendelian randomization (MR) analysis was performed to elucidate the causal connection between AITD and 41 inflammatory cytokines. Genetic variations associated with inflammatory cytokines were sourced from the FinnGen biobank, whereas a comprehensive meta-analysis of genome-wide association studies (GWASs) yielded data on Graves’ disease (GD) and Hashimoto thyroiditis. Regarding the MR analysis, the inverse variance-weighted, MR-Egger, and weighted median methods were utilized. Additionally, sensitivity analysis was conducted using MR-Egger regression, MR-pleiotropy residual sum, and outliers. Results Seven causal associations were identified between inflammatory cytokines and AITD. High levels of tumor necrosis factor–β and low levels of stem cell growth factor–β were indicative of a higher risk of GD. In contrast, high levels of interleukin-12p70 (IL-12p70), IL-13, and interferon-γ and low levels of monocyte chemotactic protein–1 (MCP-1) and TNF-α suggested a higher risk of HD. Moreover, 14 causal associations were detected between AITD and inflammatory cytokines. GD increases the levels of macrophage inflammatory protein–1β, MCP-1, monokine induced by interferon-γ (MIG), interferon γ–induced protein 10 (IP-10), stromal cell–derived factor–1α, platelet-derived growth factor BB, β–nerve growth factor, IL-2ra, IL-4, and IL-17 in blood, whereas HD increases the levels of MIG, IL-2ra, IP-10, and IL-16 levels. Conclusion Our bidirectional MR analysis revealed a causal relationship between inflammatory cytokines and AITD. These findings offer valuable insights into the pathophysiological mechanisms underlying AITD.


Introduction
Autoimmune thyroid disease (AITD) encompasses a group of organ-specific autoimmune diseases characterized by the production of antithyroid antibodies and infiltration of lymphocytes in the thyroid gland (1).Epidemiological investigations established that its prevalence is approximately 5% in the general population and has steadily increased in recent years (2,3).Clinically, Graves' Disease (GD) and Hashimoto thyroiditis (HT) are the most common forms of AITD, hallmarked by hyperthyroidism and hypothyroidism, respectively (4).Among them, the former represents a specific autoimmune disease featured by increased thyroid hormone secretion driven by an immune system imbalance that promotes the production of thyroid-stimulating antibodies (5).Additionally, as an AITD, the pathogenesis of HT is similar to that of GD.Specifically, the pathogenesis of AITD involves genetic defects and genetic susceptibility, psychological factors, infections, and stress reactions.Meanwhile, environmental pollution and a high-iodine diet can contribute to or exacerbate autoimmune reactions leading to HT (6).At present, there is an urgent need to investigate the pathogenesis of AITD to develop insights into clinical interventions.
Inflammatory cytokines are a group of proteins or peptides generated by immune cells and exhibit a diverse range of biological activities.Consequently, they play a pivotal role in autoimmune responses (7,8).Previous studies have documented a close correlation between inflammatory cytokines and the development and progression of AITD (9,10).On the one hand, cytokine imbalance directly or indirectly induces thyroid epithelial cell growth, abnormal differentiation, and immune dysfunction and can also change the state of immunoreactive cells and participate in the occurrence and development of AITD (11).On the other hand, lymphocytes infiltrating the thyroid in patients with AITD can also contribute to cytokine production.For example, in the helper T cell Th1/Th2 subgroups, Th1 cells, predominantly expressing interleukin-2 (IL-2) and interferon-g (IFN-g), stimulate delayed hypersensitivity and mediate cellular immune responses.In contrast, Th2 cells, characterized by expression of IL-4, IL-5, IL-10, and IL-13, modulate antibody production and humoral immune responses (12).At the same time, a growing body of evidence has highlighted the pivotal role of Th17 cells in the development of AITD.Of note, Th17 cells predominantly synthesize IL-17A, IL-17F, IL-21, and IL-22, along with their associated cellular and secretory components (13).Although an increasing number of studies have documented an association between inflammatory cytokines and AITD, their causal relationship remains to be investigated.
Given that genetic predisposition plays a key role in the development of AITD, it should be explained from the perspective of genetics (14).Mendelian randomization (MR) is a superior analytical approach for drawing etiological conclusions by employing genetic variation as an instrumental variable (IV) for exposure.This approach has a lower susceptibility to confounding factors owing to the random allocation of germline genetic variation during meiosis, allowing it to accurately represent exposure without being influenced by reverse causation (15).Importantly, MR studies can be performed to explore numerous dependable genetic variants owing to the public release of extensive gene-wide association data.Hence, this investigation aimed to examine the causal relationship between inflammatory cytokines and AITD.

Study population
As depicted in Figure 1, MR relies on three fundamental assumptions: (1) IVs exhibit a robust association with the exposure factor; (2) IVs are not correlated with any confounding factors; and (3) IVs solely impact the results through the exposure and not via any other mechanisms (16).In the current study, pooled data from the published genome-wide association studies (GWASs) of 41 inflammatory cytokines and AITD were utilized.Firstly, genetic variants associated with each inflammatory factor were selected to identify the causal relationship of each inflammatory cytokine with GD and HT.Secondly, AITD-associated genetic variants were employed to infer causality between GD and HT and inflammatory cytokines, respectively.
As detailed in Table 1, the genetic associations of 41 inflammatory cytokines were analyzed using data from 8,293 Finnish participants in the Cardiovascular Risk in Finnish Youth Study and the FINRISK Study (17).The first step involved genetic adjustment for 10 principal genetic components, comprising age, gender, and body mass index, as well as for population stratification and cryptic kin using genomic control.The GWAS dataset associated with GD and HD was derived from the UK Biobank Project (18).In order to expand the genetic association map beyond European populations, 220 in-depth GWASs were conducted using data from the Biobank in Japan, integrating text mining of past medical histories and electronic medical records.A meta-analysis involving the UK Biobank and FinnGen yielded approximately 5,000 new loci, thereby enhancing the resolution of the human trait genome map.Finally, a statistical

Single-nucleotide polymorphism selection
In the present study, single-nucleotide polymorphisms (SNPs) that are significantly associated with the relative abundance of 41 inflammatory cytokines were selected as IVs.Prior investigations have established that incorporating multiple IVs enhances the interpretation of exposure variation, thereby enhancing the accuracy and reliability of the results.Consequently, the 41 inflammatory cytokines were analyzed on the basis of the outcomes of a significant association analysis with P < 1 × 10 −5 .The linkage disequilibrium criteria were set at r2 < 0.001, whereas the genetic distance was established as 10,000 kb.To ensure the independence of the included SNPs, highly correlated SNPs were excluded.Subsequently, SNPs associated with the relative abundance of the 41 inflammatory cytokines were integrated into the GWAS data of AITD, and corresponding statistical parameters were extracted.The data were harmonized by comparing the statistical parameters at the same sites in the GWAS results of inflammatory cytokines relative abundance and AITD, which facilitated the alignment of the effect values of exposure and outcome with the same effect allele.

Statistical analysis
In the present study, the inverse variance-weighted (IVW), MR-Egger, and weighted median (WME) methods were employed to comprehensively assess causal effects.The IVW approach assumes the validity of all genetic variants.The causal effect value of each individual IV is calculated by IVs using the ratio method.Afterward, each estimate is meticulously summarized for weighted linear regression, yielding the total effect value (19).Regarding regression, the MR-Egger method and the IVW method primarily diverge in their consideration of the intercept term (20).The WME approach utilizes the intermediate impacts of every accessible genetic variation.Estimates were acquired by weighing the inverse variance of each SNP's association with the outcome (21).
The IVW method is superior to the other two MR methods.Therefore, the IVW method was selected as the preferred method for estimating causal effects.In order to mitigate the risk of false positives, the Benjamini-Hochberg (BH) method was used to control the false discovery rate (FDR).For a clearer explanation of the findings, the impact of the 41 inflammatory cytokines on GD and HT was present as odds ratios (ORs) for each 1 standard deviation genetic predicted change in cytokine levels.In contrast, the impact of GD and HT on the levels of inflammatory cytokines was expressed as b coefficient and 95% confidence intervals (95% CI).Considering that the connection between effect estimates and testing causation might be influenced by weak instrumental biases, the F-statistic was employed to assess the strength of the IVs using the following formula: F = R2 (n − K −1)/k (1 − R2) (22), where R2 represents the variance (per circulating cytokine) explained by the independent variable (IV), and n represents the sample size.R2 was estimated on the basis of the minor allele frequency (MAF) and bvalue, calculated using the equation: R2 = 2 × MAF × (1 − MAF) × b2 (23).
Furthermore, to ensure the stability and reliability of the findings, quality control consisting of sensitivity analysis, heterogeneity test, and gene pleiotropy test were carried out at every level.To assess the impact of each SNP on the outcome heterogeneity test, sensitivity analysis was employed using the leave-one-out method, and the combined effect size of the remaining SNPs was determined through sequential deletion of individual SNPs.In the case of significant heterogeneity between IVs (Q_pval < 0.05), the MR effect size was estimated using random-effect IVW.Otherwise, fixed-effect IVW was adopted (24).In order to examine the strength of the findings, additional sensitivity analyses were performed, including MR-Egger regression, MR-pleiotropy residual sum, and outliers (MR-PRESSO) tests.Specifically, WME accurately estimates the causal impact even with less than half of the data originating from the unreliable IV (21).The P-value of the intercept term in MR-Egger regression can serve as an indication of directional pleiotropy (25).Concerning the MR-PRESSO test, the multi-effect effect was corrected by excluding the outlier.In this study, MR analysis and quality control were conducted using R version 4.0.3 and TwoSampleMR version 0.5.6.

Two-sample MR analysis
The flow chart of 41 inflammatory cytokines and AITD is shown in Figure 1.The calculated F-values ranged from 11.156 to 788.955, all meeting the threshold of greater than 10, indicating that weak instrumental bias was unlikely (Supplementary Tables 1-4).Finally, we identified 21 causal associations between inflammatory cytokines and AITD (Tables 2, 3).

Inflammatory cytokines and autoimmune thyroid disease
In the MR analysis, the level of TNF-b was positively correlated with the risk of GD (OR, 1.115; 95% CI, 1.024-1.215;P = 0.013).However, the level of stem cell growth factor-b (SCGF-b) was negatively correlated with the risk of GD (OR, 0.913; 95% CI, 0.839-0.994;P = 0.035).Interestingly, no correlation was noted between the levels of inflammatory cytokines and GD following BH correction (Figure 2; Table 2).
Furthermore, the WME method did not detect a significant correlation between the level of tumor necrosis factor-b (TNF-b) (OR, 1.096; 95% CI, 0.981-1.225;P = 0.104) and SCG-b (OR, 0.989; 95% CI, 0.872-1.121;P = 0.857) on the risk of GD, but the direction of the effect was consistent with that of IVW (Table 2; Supplementary Figure 1).As anticipated, the MR-Egger regression intercept showed no evidence of pleiotropy between TNF-b, SCGF-b, and GD (intercept P = 0.511 for TNF-b and intercept P = 0.915 for SCGF-b).No outliers were detected by MR-PRESSO regression.Heterogeneity and sensitivity analysis results corroborated the accuracy of the results (Table 4).Lastly, the leaveone-out method further validated the robustness of the data (Supplementary Figure 5).
No outliers were detected by MR-PRESSO regression.Heterogeneity analysis results confirmed the accuracy of the results (Table 4).At the same time, leave-one-out method further verified the robustness of the data (Supplementary Figure 6).

Discussion
This study utilized MR to investigate the causal connection between AITD and inflammatory cytokines.To the best of our knowledge, this is the first study to uncover a causal association between the levels of circulating cytokines and AITD.Specifically, high levels of circulating TNF-b and low levels of SCGF-b were Associations between genetically predicted inflammatory cytokines and Graves' disease.SNPs, single-nucleotide polymorphisms; OR, odds ratio; 95% CI, 95% confidence interval; red means positive correlation; green means negative correlation.associated with a higher risk of GD.TGF-b belongs to the growth and transformational growth family (26) and plays a key role in the pathogenesis of GD (27,28).In patients with GD, the TGF-bstimulated humoral response of Th2 cells to infiltrate the thyroid gland leads to the synthesis of thyroid-stimulating hormone receptor (TSH) autoantibodies (TRAb), thereby promoting the development of GD (29)(30)(31).SCGF-b plays a vital role in promoting hematopoiesis and hematopoietic recovery.In tissues and organs beyond the bone marrow, SCGF-b can rapidly activate dormant stem cells and stimulate their growth, simultaneously regulating the internal microenvironment to create favorable growth conditions for stem cells (32, 33).Therefore, we postulate that the decrease in the level of SCGF-b may be associated with the inhibition of thyroid stem cell activity in patients with GD.Concurrently, patients with high circulating levels of IL-12p70, IL-13, and IFN-g and low levels of MCP-1 and TNF-a are at a higher risk of developing HD.IL-12p70, also referred to as natural killer (NK) cell-stimulating factor, serves as a ligand for IL-12 (34).Its primary immunomodulatory function lies in inducing the differentiation of early helper T cells into Th1 cells, which, in turn, generate pro-inflammatory proteins that trigger macrophage activation, ultimately leading to the initiation of cytotoxic effects (35).Earlier studies have established that IL-12 can exacerbate cytotoxic effects in mice with autoimmune thyroiditis by eliciting a Th1 immune response (36).IL-13, a member of the interleukin family, is largely secreted by activated Th2 cells and plays a critical role in allergic inflammation (37).It promotes M2 polarization.Furthermore, IL-13 actively contributes to the survival, activation, and recruitment of eosinophils (38).Presently, studies investigating the association between IL-13 and AITD are scarce.Therefore, it is essential to undertake comprehensive studies in the future to  Associations between genetically predicted inflammatory cytokines and Hashimoto thyroiditis.SNPs, single-nucleotide polymorphisms; OR, odds ratio; 95% CI, 95% confidence interval; red means positive correlation; green means negative correlation.
Associations between genetically predicted Graves' disease and inflammatory cytokines.SNPs, single-nucleotide polymorphisms; 95% CI, 95% confidence interval; red means positive correlation.elucidate this relationship.IFN-g is the sole member of type II interferons and is produced not only by T cells, macrophages, and NK cells but also by thyroid cells (39).Its immunomodulatory effects include increased phagocytosis of macrophages, NK cell activation, enhancing the expression level of its antigens, and the toxicity of sensitized lymphocytes on target cells (9).Animal studies have documented that IFN-g-induced apoptosis of thyroid cells is implicated in the pathogenesis of HT (40).Clinical studies further unveiled a significant rise in the number of IFN-g+ cells in patients with HT (41).MCP-1 belongs to the CC chemokine family and plays a fundamental role in the inflammatory process (42).Currently, numerous clinical studies have observed an elevated expression level of MCP-1 in both thyroid tissue and serum of patients with AITD (43,44).Importantly, our study found that low levels of MCP-1 were associated with a higher risk of HD, whereas reverse MR analysis revealed up-regulated MCP-1 expression in patients with GD.TNF-a plays an instrumental role in modulating the inflammatory response, apoptosis, and immune cell activity.Additionally, it triggers the transcription and expression of a diverse array of cytokines, thereby facilitating their synthesis and release (45).Given its central role in the inflammatory response and immune regulation, researchers have developed numerous drugs targeting TNF-a for conditions such as rheumatoid arthritis (46) and Crohn's disease (47).Notably, our study detected a negative association between TNF-a levels and HD.Further investigations with more extensive databases are warranted to validate this correlation.
Reverse MR analysis showed that GD can increase the blood levels of MIP-1b, MCP-1, MIG, IP-10, SDF-1a, PDGFbb, bNGF, IL-2ra, IL-4, and IL-17, whereas HD can increase the blood levels of MIG, IL-2ra, IP-10, and IL-16.After BH correction, a strong positive correlation was identified between the risk of HD and the expression level of MIG.MIP-1b is a CC chemokine that selectively binds to the CCR5 receptor.It chemotaxes NK cells, monocytes, and various other immune cells and can be generated by mast cells, endothelial cells, macrophages, and CD8+ T cells (48).Kemp et al. (46) observed an increased expression of MIP-1b in the thyroid tissue of patients with AITD, thereby confirming the significant role of MIP-1 in pro-inflammatory processes.MIG belongs to the glutamic acid-leucine-arginine (ELR)-negative CXC chemokine subfamily and can be triggered by IFN-g (49).CXCL9 plays an instrumental role in modulating the immune system and promoting inflammation.Of note, the recruitment, transport, and maintenance of particular subgroups of activated lymphocytes are essential for initiating and sustaining AITD (50).The secretion of chemokines that bind to CXCR3 by thyroid cells is stimulated by IFN-g, subsequently attracting Th1 lymphocytes that express CXCR3 and secreting IFN-g (51).Disrupting MIG could alleviate inflammatory reactions.Therefore, we postulate that targeting MIG may constitute a therapeutic target for the treatment of AITD.IP-10 is a member of the CXC chemokine family (52) and shares receptors such as CXCR3 with MIG.Inflammatory cells, attracted by IP-10, include chemotactic T lymphocytes that infiltrate and proliferate, thereby mediating antibody-specific autoimmune responses and resulting in gland destruction (53).Romagnani et al. (54) documented that the expression of IP-10, MIG, and CXCR3 was downregulated or absent in healthy thyroid tissue, whereas chemokines and their receptors were abundant in the thyroid glands of the majority of patients with GD.IP-10 and MIG were localized in infiltrating lymphocytes, macrophages, and resident epithelial follicular cells, whereas CXCR3 was predominantly localized in infiltrating inflammatory cells.SDF-1a belongs to the a subfamily of chemokines.Acting as a ligand, it binds to the  (72).Furthermore, existing evidence suggests that the expression levels of IL-16 in the serum could be a candidate marker for assessing disease activity and the severity of AITD (73).
To sum up, cytokines are involved in the pathogenesis of AITD (74).Their secretion profile can be either pro-inflammatory or antiinflammatory and either pro-apoptotic or anti-apoptotic (75).
These cytokines can participate in governing the immune system and the differentiation, growth, and secretory function of thyroid cells through their endocrine, autocrine, and paracrine modes.By modulating TSH levels, thyroid cells induce the abnormal expression of major histocompatibility complex-II antigens, which are transformed into antigen-presenting cells and resulting in autoimmune pathological damage and the development of AITD (76).Th1 lymphocytes generate pro-inflammatory proteins, including IFN-g and IL-2, driving macrophage activation and inducing cytotoxic effects.On the other hand, Th2 lymphocytes generate anti-inflammatory proteins that suppress the production of Th1 cytokines.Additionally, they primarily stimulate B cells to generate antibodies and trigger the activation of anti-apoptotic molecules (77).Th17 lymphocytes release pro-inflammatory cytokines such as IL17 and exert a significant impact on persistent inflammation (78,79).
While this study established a causal relationship and identified candidate targets for subsequent functional studies, it also has limitations that should not be overlooked.(1) This study used European population GWAS data for MR analysis, necessitating further studies in other populations.(2) Although MR is a highly efficient method for causality analysis, future animal tests are warranted to corroborate our findings.(3) The relationship between inflammatory cytokines and AITD is multifaceted, and elucidating the etiology and pathogenesis of circulating cytokines requires exploration from multiple aspects.(4) Ascribed to the complex etiology of AITD, circulating cytokines cannot fully explain the pathogenesis, necessitating more comprehensive data for further investigation.
In summary, the present study utilized the MR approach to investigate the causal association between inflammatory cytokines and AITD.Our results collectively unveiled that high levels of TNFb and low levels of SCGF-b were associated with a high risk of developing GD.At the same time, high levels of IL-12p70, IL-13, and IFN-g and low levels of MCP-1 and TNF-a suggest a higher risk of developing HD.Moreover, GD can increase the blood levels of MIP-1b, MCP-1, MIG, IP-10, SDF-1a, PDGFbb, bNGF, IL-2ra, IL-4, and IL-17, whereas HD can lead to elevated levels of MIG, IL-2ra, IP-10, and IL-16 levels.These cytokines hold significant implications for noninvasive early diagnosis of AITD.Finally, these cytokines may represent novel targets for the prevention, treatment, and long-term management of AITD.

FIGURE 1
FIGURE 1Assumptions and study design of the MR study of the associations between 41 inflammatory factors and autoimmune thyroid disease.BMI, body mass index; IVs, instrumental variables; SNPs, single-nucleotide polymorphisms; MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier.

TABLE 1
Details of the studies and datasets used in the study.

TABLE 2
Effects of the relationship between meaningful inflammatory cytokines and autoimmune thyroid disease in MR analysis.

TABLE 3
Effects of the relationship between autoimmune thyroid disease and meaningful inflammatory cytokines in reverse MR analysis.

TABLE 4
Heterogeneity and sensitivity analysis of meaningful inflammatory cytokines and risk of autoimmune thyroid disease.

TABLE 5
Heterogeneity and sensitivity analysis of autoimmune thyroid disease with meaningful inflammatory cytokines.

TABLE 5 Continued
IL-17+, and Th17 cells in the blood of untreated patients with AITD was higher than that of healthy individuals.Additionally, El-Zawawy et al. (71) reported a significant elevation in the expression level of serum IL-17A in patients with AITD compared to the healthy control group.IL-16 is a pro-inflammatory cytokine that exerts chemotactic effects on CD4 T lymphocytes, monocytes, and eosinophils, playing a role in both the inflammatory response and tumor development.Ongoing research has provided evidence of the expression of the IL-16 protein in the thyroid glands of patients diagnosed with GD and HT FIGURE 5Associations between genetically predicted Hashimoto thyroiditis and inflammatory cytokines.SNPs, single-nucleotide polymorphisms; 95% CI, 95% confidence interval; red means positive correlation.